Is episodic-like memory like episodic memory?

The paper examines whether episodic-like memory in non-human animals shares core characteristics with human episodic memory. It frames episodic memory as the conscious, reconstructive recollection of personal events (distinct from semantic memory), tightly linked to mental time travel and dependent on hippocampal function. Examples from amnesic patients illustrate dissociations between episodic and semantic memory and shared deficits in remembering the past and imagining the future. The introduction emphasizes the reconstructive nature of episodic recall, automatic single-trial encoding, and broader cortical networks (medial temporal, prefrontal, parietal, perceptual and imagery-related regions) that modulate phenomenology (vividness, perspective, emotion). Because non-human animals cannot provide linguistic reports of conscious experience, the article adopts the term episodic-like memory to denote behavioural signatures of episodic memory without direct access to subjective experience, and sets the goal of comparing features across taxa and paradigms to assess similarities and differences with human episodic memory.

The review surveys three major behavioural paradigms used to assess episodic-like memory across taxa and synthesizes findings on whether these capture key features of human episodic memory.

• What–where–when (WWW) paradigm: Based on Tulving’s emphasis on temporally dated episodes and spatial-temporal relations, criteria include content (what, where, when), structure (integrated representation), and flexibility (information encoded without prespecified use). Foundational evidence comes from scrub-jays that cache perishable/non-perishable foods, recalling food type, cache location, and elapsed time to guide recovery. Integration of features and flexibility (integration of novel semantic information post-event) have been demonstrated in corvids; content evidence exists across birds, rodents, primates, dogs, and cephalopods, with fewer studies meeting structure and flexibility criteria.
• Incidental encoding and unexpected question: Addresses concern that explicit task training may allow non-episodic strategies. Human episodic memory automatically encodes incidental details; thus, animals are unexpectedly queried about incidentally encoded information. Evidence spans pigeons, rats, dogs, cats, corvids and dolphins, showing recall of incidentally encoded spatial, visual, social, olfactory, and self/other actions.
• Source memory: Tests memory for the origin/context of information (perceptual modality, agent, context), a hallmark of human episodic retrieval supported by source monitoring. Evidence is emerging: rhesus macaques can dissociate item from source, rats show reality monitoring (self- vs experimenter-generated), and cuttlefish differentiate acquisition modality (smell vs sight). The review then analyzes whether episodic-like memory shares specific features of human episodic memory: hippocampal dependence, age-related decline, reconstructive nature, single-trial automatic encoding, association with future planning, contextual flexibility, and potential subjective experience. It synthesizes behavioural and neurophysiological data (e.g., hippocampal place cells, sharp-wave ripples, replay) primarily from mammals, while discussing differences in avian pallial organization and cephalopod vertical lobe analogies. Overall, evidence is mixed and taxon- and paradigm-dependent, with notable gaps in non-mammalian neurophysiology, source memory, and contextual flexibility.

This is a narrative review synthesizing comparative cognition and neuroscience literature rather than reporting new experiments. The authors organize prior research into three established paradigms that behaviorally assay episodic-like memory:

• What–where–when tasks: Single-trial, trial-unique episodes assess recall of item identity (what), location (where), and temporal component (when/how long ago/which occasion). Stringent versions test integrated binding of features (structure) and flexible use of information acquired post-event (flexibility). Performance is typically measured via foraging choices, search patterns, or object exploration preferences.
• Incidental encoding with unexpected questions: Subjects experience events without any cue that a memory test will follow; later, an unanticipated probe queries details (e.g., where a response occurred, presence/absence of food, identity of an experimenter, odours). Designs preclude preselection of actions based on anticipated tests, aiming to index automatic encoding and retrieval of incidentals. Responses include report-like discriminations (peck/approach/panel choice), spatial revisits, or action recall.
• Source memory tests: After exposure to items/events in different contexts or modalities (e.g., seen vs smelled; self- vs experimenter-generated), later tests require both item recognition and attribution to the correct source (context, modality, agent). Tasks separate item memory from source memory to demonstrate source monitoring processes. Neurophysiological evidence is incorporated to interpret mechanisms: hippocampal lesion/activation studies in rodents and primates, place-cell replay and sharp-wave ripples (awake and sleep), and comparative neuroanatomy (mammalian hippocampus vs avian hippocampus with different pallial inputs; cephalopod vertical lobe). The review then evaluates seven target questions comparing episodic-like and human episodic memory: hippocampal dependence, age-related decline, reconstructive retrieval and feature binding, single-episode automatic encoding, links to future planning, contextual flexibility, and potential subjective experience. No systematic search or meta-analytic methods are reported; the synthesis draws on influential and recent studies across taxa to build converging lines of evidence.

• Hippocampal dependence: In mammals, hippocampal lesions impair episodic-like tasks (what–where–when; incidental encoding). Place cells in rodents replay past sequences during pauses and sharp-wave ripples, paralleling aspects of human episodic recall. Birds show hippocampal involvement in spatial aspects but have different pallial connectivity; cephalopods lack a hippocampus but possess a functionally analogous vertical lobe.
• Age-related decline: Human episodic memory declines with age. Parallel declines are reported in non-human mammals on related tasks (mice, dogs, marmosets/monkeys), including incidental encoding and WWW variants. Cuttlefish do not show an age-related decline in WWW performance, suggesting hippocampal involvement may underlie mammalian age effects.
• Reconstructive nature and feature binding: Evidence for source memory (macaques, rats, cuttlefish) and binding of multiple features (rats, jays, apes) supports reconstructive retrieval. Jays integrate semantic information acquired after the event with episodic-like memory during retrieval.
• Single-episode and automatic encoding: By design, episodic-like paradigms use single, unique episodes. Incidental encoding tasks demonstrate automatic encoding and later retrieval of non-signaled details. Rodent data show single-exposure learning (e.g., avoidance of a shock zone after one experience with anticipatory place-cell activity).
• Links to future planning: Evidence is mixed and debated. Strongest support comes from corvids (New Caledonian crows passing a preregistered spoon test). Apes show tool-saving/planning in some studies, while other primates and rats often fail under stricter controls; rats exhibit neural preplay toward goals suggesting precursors to planning. Cuttlefish show flexible, future-dependent foraging.
• Contextual flexibility: Many animal studies rely on food-related or ecologically typical contexts; evidence for broad, domain-general flexibility akin to humans is limited. Some non-natural tasks and incidental social information recall (e.g., dolphins recalling human experimenter identity and spatial info) suggest potential flexibility but may be influenced by extensive training histories.
• Subjective experience: Direct evidence is unavailable. However, incidental recall demands holistic retrieval and manipulation of incidentally encoded details, and source memory involves assessments tied to past perception. Neural replay at real-world speeds and coordinated hippocampal–visual cortex activity during sleep, along with animal dreaming evidence, suggest a capacity for internally generated experiences, though not definitive for conscious recollection. Overall conclusion: There is converging evidence that episodic-like memory shares several properties with human episodic memory (single-trial formation, hippocampal dependence in mammals, reconstructive elements, automatic/incidental encoding, and some association with future planning), but with notable differences and gaps, especially across non-mammalian taxa and in contextual generality.

The review addresses the central question by evaluating whether behavioural and neural signatures of episodic-like memory in non-human animals map onto defining aspects of human episodic memory. The synthesis indicates substantial overlap in mammals: hippocampal necessity, temporal decay patterns with retention intervals, reconstruction via feature binding and source monitoring, and automatic encoding. This suggests that core mechanisms supporting event memory are evolutionarily conserved. However, divergence in avian pallial inputs to the hippocampus and cephalopod reliance on a vertical lobe imply that ostensibly similar behaviours may rest on different neural architectures and potentially yield phenomenologically different recall experiences. Mixed results and debates around animal future planning temper strong claims of equivalence. The limited contextual flexibility observed in most animal studies contrasts with the domain-generality of human episodic memory, potentially reflecting methodological biases toward ecologically naturalistic tasks but also possibly indicating real constraints on how episodic-like memory is used outside specialized behaviours. The review underscores that no single paradigm can conclusively demonstrate episodic-like memory; instead, consistent performance across multiple paradigms strengthens inferences. Incorporating neurophysiological evidence (e.g., place-cell replay, sharp-wave ripples) and expanding comparative work beyond mammals are essential to clarify homology versus convergence and to probe the presence of subjective experience during recall.

The paper concludes that episodic-like memory shares several key features with human episodic memory—formation from single episodes, dependence on the hippocampus in mammals, age-related decline (in mammals), reconstructive retrieval with feature binding and source monitoring, automatic/incidental encoding, and potential links to future planning—while differing across taxa in neural substrates and possibly in phenomenology. Evidence for broad contextual flexibility is limited, and direct evidence for subjective experience remains elusive. The authors advocate for an episodic-like memory test battery combining what–where–when (including structure and flexibility criteria), incidental encoding/unexpected question, and source memory paradigms to obtain converging evidence and reduce alternative non-episodic explanations. They call for expanded neurophysiological investigations across diverse taxa (including birds and cephalopods), and for studies in wild populations using controlled experimental approaches to better understand ecological validity and evolution of episodic-like memory. Future work should probe contextual limits of both animal episodic-like and human episodic memory and further elucidate neural mechanisms such as place-cell replay and sharp-wave ripples across species.

• Inference from behaviour: Without linguistic reports, subjective experience during recall cannot be directly assessed; all paradigms rely on excluding non-episodic strategies, which may not be exhaustive.
• Paradigm constraints: Training and task structures can allow alternative mechanisms (associative rules, preselected actions) to solve tasks, particularly when encoding is anticipated. Structure and flexibility criteria in WWW tasks are less frequently satisfied than content.
• Taxonomic gaps: Neurophysiological evidence is concentrated in rodents; there is limited comparative data in birds and cephalopods. Source memory and incidental encoding studies remain sparse across many taxa.
• Contextual bias: Many studies emphasize food-related or ecologically typical contexts, limiting conclusions about domain-generality. Captive settings and human-raised subjects may not reflect natural conditions.
• Future planning controversies: Mixed behavioural findings and debates about appropriate controls complicate interpretation of links between episodic-like memory and foresight.
• Review scope: Narrative synthesis without systematic search or meta-analysis; potential selection bias and lack of quantitative effect estimation.